A New Approach to Automated Cold Pool Tracking

Convective cold pools have generated considerable interest due to their role in modulating the organization of individual storm cells into larger systems. It is known that cold pools assist in generating new convection, but the processes by which they do so are not yet fully understood. One approach to investigate convective cold pools and the processes by which they modulate convective organization is to track individual cold pools over their respective lifetimes. However, cold pool tracking is not a straightforward task, even in numerical simulations where all state variables, forcing terms, and time tendencies are known quantities. New cold pools often overlap with pre-existing cold pools, limiting the success of recent attempts to apply watershed segmentation techniques to delineate cold pool regions.

In this work, a novel approach to automated cold pool detection and tracking is developed using a combination of existing tracking algorithms and image segmentation techniques that have previously received limited attention. The resulting algorithm is applied to cloud-resolving model simulations of tropical oceanic convection in radiative-convective equilibrium.